http://www.diva-portal.org

This is the published version of a paper published in BMC Medical Genetics.

Citation for the original published paper (version of record):

Henmyr, V., Carlberg, D., Manderstedt, E., Lind-Halldén, C., Säll, T. et al. (2017) Genetic variation of the toll-like receptors in a Swedish allergic rhinitis case population. BMC Medical Genetics, 18(1): 18 https://doi.org/10.1186/s12881-017-0379-6

Access to the published version may require subscription.

N.B. When citing this work, cite the original published paper.

Permanent link to this version: http://urn.kb.se/resolve?urn=urn:nbn:se:hkr:diva-16592 Henmyr et al. BMC Medical Genetics (2017) 18:18 DOI 10.1186/s12881-017-0379-6

RESEARCHARTICLE Open Access Genetic variation of the Toll-like receptors in a Swedish allergic rhinitis case population V. Henmyr1,2*†, D. Carlberg2†, E. Manderstedt1,2, C. Lind-Halldén2, T. Säll1, L. O. Cardell3 and C. Halldén2

Abstract Background: Variation in the 10 toll-like (TLR) has been significantly associated with allergic rhinitis (AR) in several candidate studies and three large genome-wide association studies. These have all investigated common variants, but no investigations for rare variants (MAF ≤ 1%) have been made in AR. The present study aims to describe the genetic variation of the promoter and coding sequences of the 10 TLR genes in 288 AR patients. Methods: Sanger sequencing and Ion Torrent next-generation sequencing was used to identify polymorphisms in a Swedish AR population and these were subsequently compared and evaluated using 1000Genomes and Exome Aggregation Consortium (ExAC) data. Results: The overall level of genetic variation was clearly different among the 10 TLR genes. The TLR10-TLR1-TLR6 was the most variable, while the TLR7-TLR8 locus was consistently showing a much lower level of variation. The AR patients had a total of 37 promoter polymorphisms with 14 rare (MAF ≤ 1%) and 14 AR-specific polymorphisms. These numbers were highly similar when comparing the AR and the European part of the 1000Genomes populations, with the exception of TLR10 where a significant (P = 0.00009) accumulation of polymorphisms were identified. The coding sequences had a total of 119 polymorphisms, 68 were rare and 43 were not present in the European part of the 1000Genomes population. Comparing the numbers of rare and AR-specific SNPs in the patients with the European part of the 1000Genomes population it was seen that the numbers were quite similar both for individual genes and for the sum of all 10 genes. However, TLR1, TLR5, TLR7 and TLR9 showed a significant excess of rare variants in the AR population when compared to the non-Finnish European part of ExAC. In particular the TLR1 S324* nonsense mutation was clearly overrepresented in the AR population. Conclusions: Most TLR genes showed a similar level of variation between AR patients and public databases, but a significant excess of rare variants in AR patients were detected in TLR1, TLR5, TLR7, TLR9 and TLR10.Thisfurther emphasizes the frequently reproduced TLR10-TLR1-TLR6 locus as being involved in the pathogenesis of allergic rhinitis. Keywords: Allergic rhinitis, Mutation spectrum, Next-generation sequencing, Rare variants, Toll-like receptor

* Correspondence: [email protected] †Equal contributors 1Department of Biology, Lund University, Lund, Sweden 2Department of Biomedicine, Kristianstad University, 291 39 Kristianstad, Sweden Full list of author information is available at the end of the article

© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Henmyr et al. BMC Medical Genetics (2017) 18:18 Page 2 of 10

Background Methods Allergic rhinitis (AR) is a global illness with a well- The present study aims to describe the genetic variation recognized impact on quality of life and work performance. of the promoter and coding sequences of the 10 TLR It is characterized by nasal obstruction, secretion and genes in a Swedish AR population of 288 patients. As itching and is often associated with symptoms of the eyes, the first larger resequencing effort in AR, this could fatigue and asthma [1]. The heritability of AR has been esti- identify rare variants that are increasing risk for AR and mated to 0.66–0.78 [2, 3] and its development is generally identify potential targets for future studies. believed to be the result of an interaction between genetical and environmental factors. Toll-like receptors (TLRs) are a Study samples well-established group of pattern-recognition receptors The 288 AR patients (140 females, 148 males, mean age encompassing ten members which initiate innate immune 33 years) were recruited at Malmö University Hospital responses [4]. Contrary to this, TLR10 has recently been (Malmö, Sweden) between the years 2003 and 2009 and shown to have inhibitory effects by down-regulating TLR2 consist of unrelated individuals from the general popula- mediated immune response [5]. The TLRs are an important tion. All patients were of Caucasian origin, with both barrier between environment and the organism and have parents born in Sweden. They were patients at the been suggested to play a key role in the development of allergy clinic and were diagnosed with symptomatic allergic disease in growing infants [6]. birch and/or timothy grass pollen induced intermittent The 10 TLR genes are located on five different chromo- AR (for more details see Additional file 1a). As described somes. TLR1, TLR6 and TLR10 arelocatedina90kbp in Nilsson et al. 2012, diagnostic procedures for the region on 4, TLR7 and TLR8 are located in a study population included face-to-face personal inter- 57 kbp region on the X chromosome, whereas the view of medical history and skin prick tests (SPT) [22] remaining genes are located in separate loci on chromo- or Phadiatop tests with at least a class two response to some 1 (TLR5), chromosome 3 (TLR9), birch and/or timothy grass pollen. A total of 59% of the (TLR2 and TLR3) and chromosome 9 (TLR4). Earlier patients showed a positive SPT for both allergens. SPT studies of genetic variation have reported a large number were performed with a standard panel of 11 common of polymorphisms for all 10 TLR genes [7, 8]. More airborne allergens (ALK-Abelló, Hörsholm, Denmark) recently, the 1000Genomes project has re-sequenced (for more details see Additional file 1b). SPT were 1092 healthy individuals from 14 populations using performed on the volar side of the forearm with saline low-coverage whole-genome and high-coverage exome buffer as negative and histamine chloride (10 mg/ml) as sequencing. In concordance with previous studies, the positive controls. A wheal reaction diameter of ≥3mm 1000Genomes populations also show high levels of was considered a positive SPT response. Atopy is de- variation in the TLR genes [9]. fined as a positive SPT reaction to at least one of the The TLR genes have been significantly associated with tested allergens and AR is diagnosed based on the pres- various allergic phenotypes in linkage and candidate ence of atopic status and typical AR symptoms as gene studies [10–14]. Single nucleotide polymorphisms defined by the Allergic Rhinitis Impact on Asthma (SNP) in TLR7 and TLR8 have been found to be associ- (ARIA) guidelines. None of the patients suffered from ated with AR in two independent studies [15, 16]. Three severe asthma and less than 10% had moderate asthma meta-genome-wide association studies (GWAS) have an- with continuous medication. Genomic DNA was isolated alyzed various allergic phenotypes including self- from blood collected in EDTA using the QIAmp DNA reported AR and allergic sensitization [17–19]. A total Blood (Qiagen, Hilden, Germany) and DNA concen- of 37 loci were associated with allergic disease and four trations were determined by fluorometry using Pico- of these loci were reported by all three studies. The Green (Molecular Probes, Eugene, OR, USA). TLR10-TLR1-TLR6 locus was one of these 4 loci. Follow- ing these studies, a replication attempt of the SNPs iden- Sanger sequencing tified in the three meta-GWAS was performed using Primers were designed using NCBI Primer-BLAST both the original phenotype definitions and a more strict (http://www.ncbi.nlm.nih.gov/tools/primer-blast/) to amp- AR definition [20]. A total of eight loci were successfully lify at least 50 bp downstream and 500 bp upstream of the replicated. The TLR10-TLR1-TLR6 locus was replicated start of exon 1 of all 10 TLR genes (for exact coordinates using three index SNPs and all phenotype definitions, and primer sequences see Additional files 2 and 3). Big including a more strictly defined AR phenotype. Also, an Dye Terminator Sanger sequencing was performed in additional meta-GWAS that analyzed a combined both directions using a 3130XLGenetic Analyzer (Applied phenotype of asthma and hay fever identified a total of Biosystems, Foster City, CA, USA). Sequences were inter- 10 loci, including the TLR10-TLR1-TLR6 locus, as asso- preted and all polymorphisms were identified using Seq- ciated with the disease [21]. Scape ver. 2.5 (Applied Biosystems) and confirmed by Henmyr et al. BMC Medical Genetics (2017) 18:18 Page 3 of 10

manual inspection. Further information about Sanger populations and the three statistics were summarized for sequencing can be found in Additional file 4. each individual gene and for the sum of all genes. One- sided tests were used since it was the possible accumula- Ion Torrent sequencing tion of rare variants in the AR-population that was The primer sets used in this study were obtained from under investigation. The permutation test randomized Ion AmpliSeq™ Designer (http://www.ampliseq.com, the alleles among patients (AR population) and controls pipeline version 2.0.3). A total of 204 systems were (EUR population) for each variable site and each of the designed covering 98.8% of the coding sequence of three statistics were calculated 100 000 times. The P- TLR1-TLR10 (for exact coordinates and primer se- values of the tests were equal to the proportions of quences see Additional files 2 and 5). Template DNA simulations where the randomized AR population had a was pooled such that each pool contained equimolar higher value than the value of the actual AR population. amounts of DNA from 12 individuals, producing a total A more detailed description can be found in Additional of 24 pools for the 288 AR patients. Next-generation file 4. Data from the Exome Aggregation Consortium sequencing (NGS) was performed using an AmpliSeq (ExAC) (Cambridge, MA (URL: http://exac.broadinstitu- strategy on an Ion Torrent PGM platform (Life Tech- te.org) [date (03, 2015) accessed]) were also used to test nologies, Carlsbad, CA, USA). The sequences were for accumulation of rare variants in the coding regions aligned against the human reference sequence (build of the ten TLR genes. The non-Finnish European popu- GRCh37) using Torrent Suite 3.6 and primer sequences lation of ExAC consists of > 30.000 individuals and were were trimmed away. Variant calling was then performed used for the extraction of polymorphisms (excluding using variant calling parameters tuned for high sensitiv- indels) present in the coding region of the ten TLRs. A ity. Annotation of the variant SNPs was accomplished by one-sided simulation test was used to investigate for the submitting them to SeattleSeq Annotation 137 (http:// probability of an excess of rare variants in the AR snp.gs.washington.edu/SeattleSeqAnnotation137/). Fur- patients compared to ExAC data. ther information about Ion Torrent sequencing can be Polymorphisms where at least one of the AR and EUR found in Additional file 4. populations showed a MAF ≥ 0.05 were also tested for association with AR using the normal approximation Genetic analysis test. For detailed descriptions of bioinformatics and gen- Publically available information on the polymorphisms etic analysis see Additional file 4, and for an overview of in the 10 TLR genes were extracted from dbSNP the study outline see Additional file 6. (http://www.ncbi.nlm.nih.gov/SNP/) and from the In- tegrated Variant Set of the 1000Genomes Project (http://ftp.1000genomes.ebi.ac.uk/vol1/ftp/release/2011 Results 0521/) release April 2012 [date (09, 2014) accessed]. Screening for polymorphisms in the promoters of the 10 The 1000Genomes data set was then subdivided into TLR genes four separate populations; individuals of European The 288 AR patients were screened for polymorphisms (EUR; 379 individuals), African (AFR; 246), Asian in the putative promoter regions of the 10 TLR genes (ASN; 286) and South American origin (AMR; 181). using Sanger sequencing in both directions. The Allele frequencies were calculated for each variation promoter regions were defined as the region 50 bp using the same allele as referent for all populations. downstream to 500 bp upstream of the start of exon 1 of Missense mutations identified in the study population all the genes. The sequence data was generally of high and in the 1000Genomes population were investigated quality with > 95% of bases having a Phred score of 30 using SIFT [23] and PolyPhen-2 [24]. or higher in > 95% of individuals. TLR2 and TLR5 had Three different statistics describing the spectrum of slightly lower quality scores with > 90% of bases having a variation were calculated to investigate for the accumu- Phred score of 20 or higher in > 75% of individuals. A lation of rare TLR variants in AR patients. The first sta- total of 37 polymorphisms were detected, 25 of these tistic calculated the number of sites where the minor were present in dbSNP and 12 were not (Table 1 and for allele frequencies (MAF) were ≤ 1% in patients (AR complete table see Additional file 7). The 10 genes population) and controls (EUR population). The second varied drastically with respect to the number of detected statistic calculated the number of variants that were polymorphisms with TLR10 harbouring the highest unique to either AR patients or EUR controls and the number of variants (13), equalling 35% of all detected third statistic compared SNPs detected in patients and polymorphisms. TLR1 and TLR6 which both reside in controls using information obtained in SIFT and the same locus as TLR10 also show higher than average PolyPhen-2 analysis. A one-sided permutation test was (3.7) numbers of polymorphisms (5 polymorphisms used to test all three statistics for equality of the each). TLR7 and TLR8 which are located in a small Henmyr et al. BMC Medical Genetics (2017) 18:18 Page 4 of 10

Table 1 Number of polymorphisms detected in the promotor already at this level the hypothesis of a strong general ac- regions of the TLR1-TLR10 genes. Promoter region is for both cumulation of rare variants in the AR patients can be populations defined as 50 bp downstream and 500 bp rejected. This is largely supported by the results from upstream of the start of exon 1. The AR population contains 288 the permutation test (Table 2). In TLR10 there is both a individuals and the EUR population contains 379 individuals higher number of SNPs with MAFs ≤ 1% and a higher AR population EUR population number of population-specific SNPs in AR patients. Gene Total ≤1%a AR-specificb Total ≤ 1%a EUR-specificc This yields a P-value of 0.00009 for AR-specific SNPs TLR1 51 1 40 0 that is sufficiently low to pass a Bonferroni correction TLR2 31 0 63 3 (Table 2). It should be noted that since the two catego- TLR3 43 3 11 0 ries are to a large extent overlapping, the tests are highly dependent. Thus, a Bonferroni correction based TLR4 00 0 22 2 on 20 tests would be very conservative. In addition, the TLR5 20 0 42 2 test of the total number of AR-specific SNPs yields an TLR6 51 1 94 5 uncorrected P-value of 0.03 in TLR3. TLR7 11 1 21 2 TLR8 20 0 31 1 Screening for polymorphisms in the coding sequences of TLR9 21 2 21 2 the 10 TLR genes The Ion Torrent sequencing of the 288 AR patients TLR10 13 6 6 7 0 0 identified a total of 119 polymorphisms in the coding Total 37 14 14 40 15 17 sequences of the 10 TLR genes, 68 with an allele a Polymorphisms with minor allele frequencies ≤ 1% frequency ≤ 1%, 43 not present in the EUR population bPolymorphisms present in the AR population and not present in the EUR population and 21 not previously described in dbSNP (Table 3 and cPolymorphisms present in the EUR population and not present in the see Additional file 8 for a complete list of variants). AR population Similarly to what was found for the promoter sequences, the coding sequences differ strongly with respect to the region on the X chromosome, both show lower numbers number of polymorphisms; also in this case TLR10 of polymorphisms (1 and 2 polymorphisms, respectively). shows the highest number (18) with TLR1 (14) and The corresponding sequence data from 379 individuals TLR6 (11) again showing high numbers of polymor- with European ancestry (EUR population) were extracted phisms. Also in this case TLR7 (8) and TLR8 (9) showed from the 1000Genomes project and compared with data a lower than average (11.9) number of variable sites. from the AR patients (Table 1 and for exact coordinates A comparison of the results of the AR and EUR popu- used see Additional file 2). The total number of poly- lations shows that both the total number of polymor- morphisms per gene was similar in the two populations, phisms and the number of rare polymorphisms are with a correlation coefficient of 0.6. A major part of this slightly higher in the EUR population, 119 versus 142 similarity is made up of the 23 polymorphisms that are and 68 versus 90 (Table 3). The distribution of the indi- in common to the AR and EUR populations. These poly- vidual polymorphisms over the TLR genes is very similar morphisms have in general high MAFs; 15 SNPs > 0.05, between the AR and EUR populations, with a correlation 7 SNPs 0.01–0.05 and 1 SNP < 0.01 (Additional file 7) and since both populations are samples from European Table 2 Probability (Psim) for accumulation of polymorphisms populations this is not unexpected. None of the promoter in AR population relative to EUR population P ≤ P SNPs with MAF ≥ 5% showed any significant deviation Gene sim ( 1%) sim (AR-specific) from Hardy-Weinberg equilibrium. When investigating TLR1 0.42 0.42 these SNPs for allele frequency differences between the TLR2 0.99 1.00 AR and EUR populations, only one polymorphism TLR3 0.28 0.03 (rs3764879) in TLR8 with an allele frequency difference of TLR4 1.00 1.00 0.09 yielded an uncorrected P-value < 0.05. The number of TLR5 1.00 1.00 rare (MAF ≤ 1%) and population-specific variants do not show a positive correlation between the populations but TLR6 1.00 0.82 they are on the other hand so few that the correlation TLR7 0.98 0.19 coefficient is uninformative. Two things are directly TLR8 1.00 1.00 apparent from Table 1, the total numbers of rare and TLR9 0.90 0.18 population-specific variants are relatively low and the TLR10 0.02 0.00009 sums of the different categories are very similar when Sum of all SNPs 0.94 0.004 comparing the AR and the EUR populations. Thus, Henmyr ta.BCMdclGenetics Medical BMC al. et (2017)18:18 Table 3 Number of polymorphisms detected in the coding regions of TLR1-TLR10. The AR population contains 288 individuals and the EUR population contains 379 individuals AR population EUR population Gene Base-pairs Total ≤ 1%a AR-specificb Newc Missense Nonsense Synonymous Damagingd Total ≤ 1%a EUR-specifice Missense Nonsense Synonymous Damagingd TLR1 2358 14 8 4 1 9 1 4 3 19 13 9 14 0 5 6 TLR2 2352 10 5 4 1 6 1 3 3 10 5 4 4 0 6 2 TLR3 2712 8 6 4 4 6 0 2 4 12 10 8 7 0 5 4 TLR4 2517 11 9 5 2 5 0 6 2 13 11 7 9 0 4 2 TLR5 2574 15 9 4 1 10 1 4 1 17 11 6 11 2 4 2 TLR6 2388 11 6 5 3 7 0 4 1 16 11 10 12 0 4 3 TLR7 3147 8 5 4 1 5 0 3 1 5 2 1 3 0 2 1 TLR8 3123 9 3 2 2 2 0 7 0 9 2 2 1 0 8 0 TLR9 3096 15 13 9 5 6 0 9 3 15 13 9 6 1 8 2 TLR10 2433 18 4 2 1 13 0 5 3 26 12 10 16 1 9 4 Total 2358 119 68 43 21 69 3 47 21 142 90 66 83 4 55 26 aPolymorphisms with minor allele frequencies ≤ 1% bPolymorphisms present in the AR population and not present in the EUR population cPolymorphisms not present in dbSNP dConcordant classification according to SIFT and PolyPhen-2 ePolymorphisms present in the EUR population and not present in the AR population ae5o 10 of 5 Page Henmyr et al. BMC Medical Genetics (2017) 18:18 Page 6 of 10

coefficient of 0.89. Looking at the rare SNPs there is still in the AR population and none in the EUR population. In a positive correlation between the two populations. TLR5 the estimated allele frequencies of the R392* muta- Comparing the numbers of rare and AR-specific SNPs it tion were similar at 0.059 and 0.060 in the AR and EUR is seen that the numbers are quite similar both for indi- populations, respectively. In order to evaluate a potential vidual genes and for the sum of all 10 genes. When a accumulation of damaging variants, a permutation test permutation test was applied to test for accumulation of based on the SIFT and PolyPhen-2 scores were used. The variants in the AR population, only the test for AR- one-sided test showed no significant differences between specific polymorphisms in TLR7 was significant with an the AR and EUR populations. uncorrected P-value of 0.02 (Table 4). Also the overall SNPs with MAF ≥ 5 were also tested individually for test of the AR-specific polymorphisms gave a significant allele frequency differences between the AR and EUR pop- result (P = 0.04) out of a total of 22 tests performed. ulations. Two SNPs showed significant allele frequency The non- Finnish European population of ExAC were differences and were located in TLR5 (rs2072493) and in used as an additional control population. Due to the dif- TLR8 (rs3764880), respectively. Both were missense mu- ferent structure of the data it was used in a simulation tations classified as “benign” by PolyPhen-2. The allele test to investigate the probability for an accumulation of frequency differences were 0.056 and 0.086 correspond- rare variants (MAF ≤ 1%) in the AR population. As in ing to P-values of 0.04 and 0.008, respectively. In both the test for AR-specific variants vs the EUR population, cases, the AR population had a lower frequency of the TLR7 again shows a significant result and yields an rare allele. uncorrected P-value of 0.0034. In addition, TLR1, TLR5 and TLR9 also gave rise to significant results with TLR9 Discussion showing the lowest P-value, indicating an excess of rare We have adopted a broad perspective in the present study variants in the AR population compared to ExAC data. where we investigate all 10 TLR genes for both rare The analysis above covered all variants. An obvious alter- (MAF ≤ 1%) and common (MAF > 1%) polymorphisms in native is to focus on variants that alter the gene products both promoters and coding sequences. The polymorphisms i.e. 3 nonsense mutations and 69 missense mutations. are also subdivided into different classes e.g. nonsense and Forty-four out of the 69 missense mutations had MAFs ≤ missense where missense mutations are further evaluated 0.01. TLR1, TLR5 and TLR10 had the highest numbers of based on functional predictions. The primary objective of non-synonymous variants with 10, 11 and 13 variants, re- the present study was to describe the genetic variation of spectively. The 3 nonsense mutations were S324* in TLR1, theTLRgenesinaSwedishARpopulationincomparison R447* in TLR2 and R392* in TLR5 (Additional file 8). The to the general variation seeninbackgroundpopulations nonsense mutation S324* in TLR1 had an allele frequency similar to our own. A second objective was to identify rare of 0.009 in the AR population (estimated to represent 5 candidate mutations that are increasing risk for AR. The copies), whereas it was not present at all in the EUR popu- main results of this study is summarized in Fig. 1. lation or the non-Finnish population of ExAC (investigating In the present study different sequencing methods 66 616 ). The corresponding frequencies for were used to create the different datasets. Since different the R447* mutation in TLR2 was estimated to a single copy sequencing platforms have different error profiles, direct comparisons can be problematic. The TLR exome data Table 4 Probability (Psim) for accumulation of polymorphisms analyzed in this study have high coverage (>50X) in in AR population relative to EUR population and the non- 1000Genomes and ExAC data, and in our Ion Torrent Finnish European population of ExAC sequencing data (~200X), for more information see P ≤ P P ≤ Gene sim ( 1%) sim (AR-specific) sim (ExAC 1%) [Additional file 9]. In addition, we have chosen not to TLR1 0.91 0.63 0.012 report any indels for the TLR exome as Ion Torrent se- TLR2 0.51 0.25 0.62 quencing is notorious for its high error rate with respect TLR3 0.76 0.79 0.32 to indels. Thus, the probability of errors is expected to be low in the different datasets. Sanger sequencing of TLR4 0.80 0.24 0.37 the promoter regions generally yielded high quality data TLR5 0.71 0.18 0.042 and the identified polymorphisms are likely to be true. TLR6 0.97 0.56 0.75 Two exceptions exist, both TLR2 and TLR5 show slightly TLR7 0.25 0.020 0.0034 lower quality, but since these two genes do not give rise TLR8 0.68 0.14 0.91 to any significant signals these genes were not evaluated – TLR9 0.53 0.16 0.00086 further. The low coverage (2 6X) NGS data from the 1000Genomes project are more likely to create both TLR10 0.91 0.91 1.00 false positive and false negative data. However, since Sum of all SNPs 0.91 0.04 0.42 both the total number and the number of rare Henmyr et al. BMC Medical Genetics (2017) 18:18 Page 7 of 10

Fig. 1 Results summary for the TLR genes. Significant differences between AR patients and 1000Genomes data are indicated with diagonal lines going from top left to bottom right. Significant differences between AR patients and ExAC data are indicated with diagonal lines going from top right to bottom left. Nonsense mutations are denoted with “STOP”. Truncated introns are indicated with gaps and the corresponding sequence lengths are given polymorphism are highly similar for the two popula- consistently showing a lower level of variation. These tions, this seems not to be a large problem. Another differences may be due to several different factors. It concern is the use of publicly available background pop- should be remembered that even under strict neutrality ulations as control groups since AR is a common disease the level of variation is expected to vary between with a prevalence of 10–20% [1]. However, the presence regions, i.e. the level of variation observed may be due of AR cases among the controls only lower the power of to chance events. X-linked loci have slightly lower effect- the analysis. This is true, both for analysis of allele fre- ive population size which may have contributed to the quencies of common variants as well as the analysis of lower level of variation in TLR7 and TLR8. In addition, rare variants. Another concern related to this is the issue since active genes are involved it can safely be assumed of population specific variants with increased allele fre- that purifying selection is operating on nonsense and quencies. But since comparisons and tests are made some of the missense mutations. If the effects of these using the collective number of alternative alleles from all changes are recessive, the effect of selection is stronger rare variants in a specific region, and not the frequencies on X-linked loci, thus further reducing the level of vari- of single variants, the effect of population specific vari- ation. This effect will be present for all sites in linkage ants will be less pronounced. disequilibrium with the selected loci. A general observation that is not directly related to Onlyahandfulofthemorefrequentpolymorphisms the occurrence of AR is the fact that the overall level of (MAF ≥ 5%) that could be tested individually yielded signifi- genetic variation is clearly different among the 10 TLR cant differences between the two investigated populations. genes and that this to a large extent is related to their SNPs in the promoter and the coding sequence of TLR8 chromosomal locations. The TLR10 gene is the most showed significant frequency differences between the AR variable in both populations, in the promoter as well as and the EUR population. These SNPs are all located in the in coding regions. The other two genes in the same same haploblock and reflects the association observed earl- region, TLR1 and TLR6, are also well above average with ier for the TLR7-TLR8 region [16]. However, no signs of any regard to the number of variable sites. The other accumulation of rare variants was detected in TLR8 which extreme is represented by TLR7 and TLR8 which are was also demonstrated in a previous study [25]. Henmyr et al. BMC Medical Genetics (2017) 18:18 Page 8 of 10

Many of the missense variants in the coding sequence previously shown to be associated with allergic diseases, may well be damaging and result in lowered fitness. but further studies investigating genetic variation in However, it is likely that a majority of these are not in- these genes and its contribution to allergic diseases are volved in the development of AR since approximately needed. equal numbers were found in cases and controls. The 3 Three large meta-GWAS identified a total of 37 loci nonsense polymorphisms (TLR1 S324*, TLR2 R447* and associated with AR where four of these loci were TLR5 R392*) are located in the leucine-rich repeats and reported by all three studies [17–19]. One of these four truncates the receptors so that neither the transmem- loci was the TLR10-TLR1-TLR6 locus which was identi- brane nor the toll-IL1 receptor domain will be translated fied by a large number of SNPs. Following these studies, and are therefore damaging to the function of the pro- a replication attempt of the SNPs identified in the three teins. The TLR1 S324* nonsense mutation was clearly meta-GWAS was performed using both the original overrepresented in the AR population compared to the phenotype definitions and a more strict AR definition EUR population and the non-Finish European ExAC [20]. A total of eight loci were successfully replicated population. The frequency observed in AR patients for including the 90 kbp TLR10-TLR1-TLR6 locus. This S324* is 45 times higher than the most common non- locus produced significant P-values for SNPs located in sense mutation in TLR1 in the complete ExAC database. a region that were larger than 100 kbp and encompassed Thus, the TLR1 S324* mutation is a strong candidate for all three genes. Another meta-GWAS investigating a being associated with AR. combined phenotype of asthma and hay fever also Investigation of rare variants in the promoter regions identified this locus as associated with the disease [21]. showed no general accumulation in AR patients relative In addition, SNPs in this region have earlier been associ- to EUR controls. The major exception was the promoter ated with altered expression levels of these genes [26]. region of TLR10 where a surplus of 6 polymorphisms The patterns of eQTL data for TLR10, TLR1 and TLR6 were identified in AR patients compared to controls. expression are shown in Additional file 10 together with This difference was highly significant. In addition, SNPs the AR-associated SNPs from the meta-GWAS of Hinds in this region have previously been associated with et al. [19]. If the pattern of these signals are compared altered expression of the TLR10 gene [26], suggesting with the corresponding pattern from the AR association that altered gene expression of TLR10 may be involved in study of Hinds et al. [19] there is a striking similarity, the pathogenesis of allergic disease. Recently it was shown indicating that the expression differences in the TLR10- that TLR10 is a receptor with suppressive effects, reducing TLR1-TLR6 region is correlated with AR. The overrepre- TLR2-mediated production including IL-1β,IL- sentation in the AR population of a set of variants in the 6, IL-8 and TNF-α. This is believed to be achieved promoter of TLR10 and in the coding region of TLR1 in- through either or all of the 3 following mechanisms; 1) cluding the S324* mutation, identify these polymor- competitive binding of ligands of stimulatory TLRs 2) phisms as candidate mutations and add further details to TLR10 competing for hetero-dimerization of other TLR2 the associations observed for common SNP from this family members (TLR1 and TLR6) 3) direct inhibitory region in the meta-GWAS and the replication study. effects of TLR10 via P13k signaling pathway [5]. The lack of significant associations of common variants The analysis of the coding sequence showed no strong in the TLR10-TLR1-TLR6 locus is probably a matter of general accumulation of rare or AR-specific polymor- power in the present study. There are in fact allele phisms in AR patients relative to EUR controls. However, frequency differences for a number of SNPs, but at a in the simulation tests for rare variants using ExAC data, non-significant level. Nonetheless, the rare variants TLR1, TLR5, TLR7 and TLR9 showed significant accumu- found in this study coupled with previous strong associ- lations of rare variants in AR patients. SNPs in the TLR1 ation to common variants highly implicate the TLR10- locus have previously been associated with allergic disease TLR1-TLR6 region as a risk-locus for AR. in a number of large meta-GWAS [17–19, 21] and in a Swedish replication study [20]. Genetic variation in Conclusions TLR7 have been associated with the development of This study was motivated by the lack of re-sequencing asthma, rhinitis, and AD in a Danish population [15] studies in allergic diseases. Previous association studies and also with skin-prick test response for house dust have identified common variants in the TLR genes to be mites in a Singaporean Chinese population [16]. In a associated with AR. The present study describes the study of Tunisian children, SNPs in TLR9 were re- genetic variation of these genes in a population selected ported to play a role in the predisposition to asthma for AR, which is an important step to understand the [14] and were also associated with asthma risk in a genetic contribution of the disease. Also, the present recent meta-study [27]. Thus, the results from the study identifies an excess of rare variants in the TLR1, simulation tests in the present study identifies genes TLR5, TLR7, TLR9 and TLR10 genes in AR patients Henmyr et al. BMC Medical Genetics (2017) 18:18 Page 9 of 10

compared to public background populations, indicating Author details 1 2 that rare variants may also contribute to the disease. Department of Biology, Lund University, Lund, Sweden. Department of Biomedicine, Kristianstad University, 291 39 Kristianstad, Sweden. 3Division of Further studies investigating the contribution of rare ENT Diseases, Department of Clinical Science, Intervention and Technology, variants in AR using larger populations consisting of Karolinska Institutet, Stockholm, Sweden. matched cases and controls are much needed, but the Received: 16 September 2016 Accepted: 8 February 2017 results of the present study may serve as a starting point for future studies.

References Additional files 1. Brozek JL, Bousquet J, Baena-Cagnani CE, Bonini S, Casale TB, van Wijk RG, et al. Allergic rhinitis and its impact on asthma (ARIA) guidelines: 2010 – Additional file 1: Demographics of study populations. (XLSX 9 kb) revision. J Allergy Clin Immunol. 2010;126:466 76. 2. Fagnani C, Annesi-Maesano I, Brescianini S, D’Ippolito C, Medda E, Nisticò L, Additional file 2: Targeted genomic regions. (XLSX 14 kb) et al. Heritability and shared genetic effects of asthma and hay fever: an Additional file 3: Sanger sequencing primers. (XLSX 9 kb) Italian study of young twins. Twin Res Hum Genet. 2008;11:121–31. Additional file 4: Materials and methods for Sanger sequencing, Ion 3. Willemsen G, van Beijsterveldt TC, van Baal CG, Postma D, Boomsma DI. Torrent sequencing, Bioinformatics analysis and Genetic analysis. (DOCX Heritability of self reported asthma and allergy: a study in adult Dutch 20 kb) twins, siblings and parents. Twin Res Hum Genet. 2008;11:132–42. 4. Pandey S, Agrawal DK. Immunobiology of Toll-like receptors: emerging Additional file 5: Ion Torrent sequencing primers. (XLSX 22 kb) trends. Immunol Cell Biol. 2006;84:333–41. Additional file 6: Outline of the study. (DOCX 12 kb) 5. Oosting M, Cheng SC, Bolscher JM, Vestering-Stenger R, Plantinga TS, Additional file 7: Characteristics of promoter polymorphisms detected Verschueren IC, et al. Human TLR10 is an anti-inflammatory pattern- in 288 AR patients and comparison with allele frequencies in recognition receptor. Proc Natl Acad Sci U S A. 2014;111(42):E4478–84. subpopulations extracted from 1000Genomes project. (XLSX 22 kb) 6. Tesse R, Pandey RC, Kabesch M. Genetic variations in toll-like receptor pathway genes influence asthma and atopy. Allergy. 2011;66:307–16. Additional file 8: Characteristics of coding sequence polymorphisms 7. Georgel P, Macquin C, Bahram S. The heterogeneous allelic repertoire of detected in 288 AR patients and comparison with allele frequencies in human Toll-like receptor (TLR) genes. PLoS ONE. 2009;4:e7803. subpopulations extracted from 1000Genomes project. (XLSX 67 kb) 8. Ma X, Liu Y, Gowen BB, Graviss EA, Clark AG, Musser JM. Full-exon Additional file 9: Ion Torrent coverage. (EPS 67 kb) resequencing reveals Toll-like receptor variants contribute to human Additional file 10: eQTL and association data for TLR10-TLR1-TLR6 locus. susceptibility to tuberculosis disease. PLoS ONE. 2007;2:e1318. (EPS 228 kb) 9. Abecasis GR, Auton A, Brooks LD, DePristo MA, Durbin RM, Handsaker RE, et al. An integrated map of genetic variation from 1,092 human genomes. Nature. 2012;491:56–65. Abbreviations 10. Medvedev AE. Toll-like receptor polymorphisms, inflammatory and AFR: African; AMR: South American; AR: Allergic rhinitis; ARIA: Allergic rhinitis infectious diseases, allergies, and cancer. J Interferon Cytokine Res. impact on asthma; ASN: Asian; EUR: European; ExAC: Exome aggregation 2013;33:467–84. consortium; GWAS: Genome-wide association study; MAF: Minor allele 11. Salpietro C, Rigoli L, Miraglia Del Giudice M, Cuppari C, Di Bella C, Salpietro A, frequency; NGS: Next-generation sequencing; SNP: Single nucleotide et al. TLR2 and TLR4 gene polymorphisms and atopic dermatitis in Italian polymorphism; SPT: Skin prick test; TLR: Toll-like receptor children: a multicenter study. Int J Immunopathol Pharmacol. 2011;24:33–40. 12. Kerkhof M, Postma DS, Brunekreef B, Reijmerink NE, Wijga AH, Funding de Jongste JC, et al. Toll-like receptor 2 and 4 genes influence This study was funded by the Swedish Research Council which only susceptibility to adverse effects of traffic-related air pollution on – provided funds and was not involved in any design or analysis of this study. childhood asthma. Thorax. 2010;65:690 7. 13. Miedema KG, Tissing WJ, Te Poele EM, Kamps WA, Alizadeh BZ, Kerkhof M, et al. Polymorphisms in the TLR6 gene associated with the inverse Availability of data and materials association between childhood acute lymphoblastic leukemia and atopic All data generated or analyzed during this study are included in this disease. Leukemia. 2012;26:1203–10. published article and its supplementary information files. 14. Lachheb J, Dhifallah IB, Chelbi H, Hamzaoui K, Hamzaoui A. Toll-like receptors and CD14 genes polymorphisms and susceptibility to asthma in Authors’ contributions Tunisian children. Tissue Antigens. 2008;71:417–25. All authors contributed to the design of the experiments. DC, VH and TS 15. Møller-Larsen S, Nyegaard M, Haagerup A, Vestbo J, Kruse TM, Borglum AD. performed statistical analysis. EM produced Ion Torrent sequencing data and Association analysis identifies TLR7 and TLR8 as novel risk genes in asthma CLH produced Sanger sequencing data. VH and DC performed and related disorders. Thorax. 2008;63:1064–9. bioinformatics analysis. LOC acquired phenotypic data and all authors 16. Nilsson D, Andiappan AK, Halldén C, De Yun W, Säll T, Tim CF, et al. Toll-like contributed to the writing of the manuscript and approved the final version receptor gene polymorphisms are associated with allergic rhinitis: a case of the manuscript. control study. BMC Med Genet. 2012;13:66. 17. Ramasamy A, Curjuric I, Coin LJ, Kumar A, McArdle WL, Imboden M, et al. A Competing interests genome-wide meta-analysis of genetic variants associated with allergic The authors declare that they have no competing interests. rhinitis and grass sensitization and their interaction with birth order. J Allergy Clin Immunol. 2011;128:996–1005. 18. Bønnelykke K, Matheson MC, Pers TH, Granell R, Strachan DP, Alves AC, et al. Consent for publication Meta-analysis of genome-wide association studies identifies ten loci Not applicable. influencing allergic sensitization. Nat Genet. 2013;45:902–6. 19. Hinds DA, McMahon G, Kiefer AK, Do CB, Eriksson N, Evans DM, et al. A Ethics approval and consent to participate genome-wide association meta-analysis of self-reported allergy identifies The present study was approved by the Ethics Committee of the Medical shared and allergy-specific susceptibility loci. Nat Genet. 2013;45:907–11. Faculty, Lund University and the Swedish Data Inspection Board and written 20. Nilsson D, Henmyr V, Halldén C, Säll T, Kull I, Wickman M, et al. Replication informed consent was obtained from all subjects. This study is also in of genome-wide associations with allergic sensitization and allergic rhinitis. compliance with the Helsinki declaration. Allergy. 2014;69:1506–14. Henmyr et al. BMC Medical Genetics (2017) 18:18 Page 10 of 10

21. Ferreira MAR, Matheson MC, Tang CS, Granell R, Ang W, Hui J, et al. Genome-wide association analysis identifies 11 risk variants associated with the asthma with hay fever phenotype. J Allergy Clin Immunol. 2014;133(6):1564–71. 22. Bousquet J, Heinzerling L, Bachert C, Papadopoulos NG, Bousquet PJ, Burney PG, et al. Practical guide to skin prick tests in allergy to aeroallergens. Allergy. 2011;67:18–24. 23. Sim NL, Kumar P, Hu J, Henikoff S, Schneider G, Ng PC. SIFT web server: predicting effects of amino acid substitutions on . Nucleic Acids Res. 2012. doi: 10.1093/nar/gks539. 24. Adzhubei IA, Schmidt S, Peshkin L, Ramensky VE, Gerasimova A, Bork P, et al. A method and server for predicting damaging missense mutations. Nat Methods. 2010;7:248–9. 25. Henmyr V, Lind-Halldén C, Carlberg D, Halldén C, Melén E, Wickman M, Bergström A, Säll T, Cardell LO. Characterization of genetic variation in TLR8 in relation to allergic rhinitis. Allergy. 2016;71(3):333–41. 26. Lonsdale J, Thomas J, Salvatore M, Phillips R, Lo E, Shad S. The Genotype-Tissue Expression (GTEx) project. Nat Genet. 2013;45:580–5. 27. Tizaoui K, Kaabachi W, Hamzaoui K, Hamzaoui A. Association of single nucleotide polymorphisms in toll-like receptor genes with asthma risk: a systematic review and meta-analysis. Allergy, Asthma Immunol Res. 2015;7(2):130–40.

Submit your next manuscript to BioMed Central and we will help you at every step:

• We accept pre-submission inquiries • Our selector tool helps you to find the most relevant journal • We provide round the clock customer support • Convenient online submission • Thorough peer review • Inclusion in PubMed and all major indexing services • Maximum visibility for your research

Submit your manuscript at www.biomedcentral.com/submit